Knapsack problem is finding the optimal selection of objects to get maximum profit. Knapsack problem has a wide range of application in different domain such as production, transportation, resource management etc. Knapsack problem varies with change in number of items and number of objectives. 01 knapsack problem is reported as a classical optimization problem under NP category. Harmony search (HS) algorithm is a popular heuristic algorithm investigated to solve different optimization problems. This paper presents harmony search for solving single objective and multi-objective knapsack problem. Performance of HS is tested with 43 instances of single objective knapsack problem taken from three datasets. HS provides optimal results except for three instances. 46 instances of 01 knapsack problem with three, four and five objectives are tested. Experiments show that better results are obtained with an increase in harmony memory with better exploration in objectives.